I think I have found a solution that achieves the desired result in good way (in terms of not writing spaghetti php or hardcoding stuff in the theming layer). Instead of providing the full configuration at once, I will try to guide you step by step.
1/ Create a new view (mine is called 'test') that only has a Block display, and displays Taxonomy terms for a specific vocabulary (mine is called 'Target Group'). Use 'HTML list' as the selected Format, make sure the
'Taxonomy term: Name' is added in the list of the displayed fields, also use 'Taxonomy term: Name' as a sorting option in ascending order, select 'Display all items' in the pager settings, and make sure query Distinct settings are DISABLED (we will achieve that via aggregation later on). Save your view. You can see the view results in the preview section of the view manage page, or place the block in the respective node display if you prefer to view it directly in the front end. At this point, you will be able to see a list of all the terms form the specific vocabulary.
2/ Add a new relationship to your view (in the Advanced section). The relationship should be the one labeled 'Taxonomy term: Content with term'. It is highly important (for the rest of the code that will be provided) that you choose this generic relationship option, instead of the more specific one labeled 'Taxonomy term: Content using Target group' (where Target Group is the vocabulary name I used for this example). Provide an identifier for this relationship - I left the default one labeled 'node'. Make sure that 'Require this relationship' is NOT checked - this will create a LEFT JOIN between the terms table and the node term data table. Now go to the fields section and add the field 'Content: Nid' and set it to use the relationship with the aforementioned identifier. Any other field settings do not matter for now - we will adjust them later on. Save your view. At this point you will be able to see plenty of duplicates for all terms, each one with a different nid. Good.
3/ Now go to the contexual filter section and add a new one for 'Content: Nid'. Make sure the 'node' relationship is used. For 'When the filter value is NOT available' choose 'Provide default value' and set 'Type' to 'Content ID from URL'. It is recommended to 'Specify validation criteria' as you see fit. Save your view. The configuration until this step, is actually the one presented in this excellent answer here which has some screenshots too. Up to this point, you will be able to see only the terms with which the current node is tagged. "Duhh, that was the original problem" you'll say, but bear with me as we take it one step further.
You see, when adding a contexual filter, views adds the restriction to the WHERE clause of the generated query (which is good and desired in the majority of cases). The current query generated is the following:
SELECT DISTINCT taxonomy_term_data.name AS taxonomy_term_data_name, taxonomy_term_data.vid AS taxonomy_term_data_vid, taxonomy_term_data.tid AS tid, taxonomy_vocabulary.machine_name AS taxonomy_vocabulary_machine_name, node_taxonomy_index.nid AS node_taxonomy_index_nid
FROM
taxonomy_term_data taxonomy_term_data
LEFT JOIN taxonomy_index taxonomy_index ON taxonomy_term_data.tid = taxonomy_index.tid
LEFT JOIN node node_taxonomy_index ON taxonomy_index.nid = node_taxonomy_index.nid
LEFT JOIN taxonomy_vocabulary taxonomy_vocabulary ON taxonomy_term_data.vid = taxonomy_vocabulary.vid
WHERE (( (node_taxonomy_index.nid = '4072' ) )AND(( (taxonomy_vocabulary.machine_name IN ('topics')) )))
ORDER BY taxonomy_term_data_name ASC
But what we DO want in our case, is that the nid restriction is applied as an additional restriction to the LEFT JOIN with the node_taxonomy_index in order to filter the join and not the whole result set. Unfortunately, views does not support that out of the box but after some digging I found that it is programmatically supported in views_join handler which can be gracefully managed via hook_views_query_alter. So, moving on.
4/ In order to implement a hook, you will require a custom module. If you are not sure how, then read the respective documentation. Here is the implementation that removes the nid restriction from the where clause and adds it to the LEFT JOIN clause
function my_custom_module_views_query_alter(&$view, &$query) {
if ($view->name == "test") {
// try to resolve the nid from the contexual filter condition
$contexual_nid = null;
if (!empty($query->where)) {
foreach ($query->where as $delta => $where_clause) {
if (isset($where_clause['conditions']) && count($where_clause['conditions'])) {
$condition = $where_clause['conditions'][0]; // in the current use case, there will be only one condition
if (isset($condition['field']) && strpos($condition['field'], "nid") !== FALSE) {
$contexual_nid = array_shift($condition['value']);
// remove the nid constraint from the where clause
unset($query->where[$delta]);
}
}
}
}
// add the nid constraint to the relationship
if ($contexual_nid != null && isset($query->table_queue['node_taxonomy_index']) && isset($query->table_queue['node_taxonomy_index']['join'])) {
$join = $query->table_queue['node_taxonomy_index']['join'];
$join->extra = array(
array(
"field" => "nid",
"value" => $contexual_nid
)
);
$query->table_queue['node_taxonomy_index']['join'] = $join;
}
}
}
So, at this point, you have a list that has tons of duplicates. "Duh, I had achieved that via 'views_field_view' and with less duplicates" you'll think. Well, no. First of all, the list is now sorted alphabetically, and second, getting rid of the duplicates, now that we are dealing with a single result set, is a piece of cake via aggregation. Back to the view administration then.
5/ In the Advanced section, set 'Use aggregation' to Yes. For field 'Content: Nid', click on the 'Aggregation settings' and set 'Aggregation type' to 'Count'. For field 'Taxonomy term: Name' the aggregation type should be the default 'Group results together'. Save your view and say goodbye to the duplicates.
6/ Last step in order to convert the 'Content: Nid' to useful markup. Click on the field which is now labeled '(node) COUNT(Content: Nid) (Nid)' and apply the following settings:
- Create a label: unckeck
- Exclude from display: check
- REWRITE RESULTS: Rewrite the output of this field: checked
- REWRITE RESULTS: Text: 'current' (the actual literal 'current'
without the quotes)
- NO RESULTS BEHAVIOR: Count the number 0 as empty: checked
- NO RESULTS BEHAVIOR: Hide if empty: checked
- NO RESULTS BEHAVIOR: Hide rewriting if empty : checked
And finally click on the Settings link for the HMTL list format and for 'Row class' enter token '[nid]'. Save your view. Everything should look as expected.
